Datasets & sinks

Arc, already in your warehouse.

Every Arc stream delivered continuously into the systems you already query. Reconciled USDC transfers, finality, logs, blocklist changes and CCTP legs, backfilled from genesis and kept current as blocks commit.

Get an API keyTalk to us
  • USDC transfers
  • Blocks + certificates
  • Logs
  • Blocklist
  • CCTP legs

  • Kafka

    topic per stream

  • Postgres

    upserts on dedup key

  • ClickHouse

    ReplacingMergeTree

  • S3

    hourly Parquet

Every range delivered comes with a signed gap report: a statement that no block in it was skipped.

Destinations

Eight sinks. Four included on Scale.

The same schema in every destination, so a query written against one works against another.

Kafka

Scale

One topic per stream, keyed on the dedup key so compaction keeps the latest record.

PostgreSQL

Scale

Typed tables with upserts on the dedup key. Point your ORM at it.

ClickHouse

Scale

ReplacingMergeTree tables ordered by block number and log index.

Amazon S3

Scale

Hourly Parquet partitions by block range, with a manifest per partition.

Snowflake

On request

Staged loads into your account, schema managed on our side.

BigQuery

On request

Streaming inserts into a dataset you own.

Redpanda

On request

Kafka-compatible delivery for teams already on Redpanda.

Kinesis

On request

One stream per Arc stream in the AWS region you choose.

Example: daily flows

Net USDC flow per address, one query.

Because every transfer is already one record with the amount at 6 decimals, a flow report is a GROUP BY. No emitter filter, no decimal conversion, no double counting to subtract back out.
clickhousesql
-- net USDC flow per address, per day (ClickHouse sink)SELECT  toDate(block_time)            AS day,  address,  sum(delta_usdc)               AS net_usdcFROM (  SELECT block_time, "to"   AS address,  amount_usdc AS delta_usdc FROM usdc_transfers  UNION ALL  SELECT block_time, "from" AS address, -amount_usdc AS delta_usdc FROM usdc_transfers)GROUP BY day, addressORDER BY day DESC, net_usdc DESC;
Example: receivables

Match payments to invoices by Memo.

The Memo contract lets a sender attach a payment reference. The transfer feed decodes it onto the transfer it belongs to, so settling receivables is a join on your own invoice table.
postgressql
-- match incoming USDC to open invoices by Memo (Postgres sink)SELECT i.invoice_id, i.amount_due, t.amount_usdc, t.block_numberFROM invoices iJOIN usdc_transfers t  ON t.memo = i.invoice_id AND t."to" = i.receiving_addressWHERE i.status = 'open';
Example: cross-chain

See USDC entering and leaving Arc.

CCTP burns and mints arrive correlated by nonce, with the remote domain attached, so treasury can see net movement per chain without stitching two contracts’ events together.
athena on s3sql
-- net USDC crossing into and out of Arc, by remote chain (S3 via Athena)SELECT  remote_domain,  sum(CASE WHEN direction = 'INBOUND'  THEN amount_usdc ELSE 0 END) AS inbound,  sum(CASE WHEN direction = 'OUTBOUND' THEN amount_usdc ELSE 0 END) AS outboundFROM cctp_legsWHERE block_time >= current_date - interval '30' dayGROUP BY remote_domain;
Guarantees

What every sink delivery carries.

01

Signed gap report

Every delivered range ships with a signed statement that no block in it was skipped. A hole is provable, not suspected.

02

Idempotent writes

Every record carries a dedup key, so a retried delivery upserts instead of duplicating.

03

Nothing to roll back

Arc finalises on commit. Sinks never receive a compensating write for a reorg, because there are none.

Setup

From credentials to a synced warehouse.

  1. Day 0

    Pick streams and a destination

    Choose the streams, filters and sink. We send the schema and the permissions the sink needs.

  2. Day 1

    Backfill from genesis

    History loads into your destination in block order, with a gap report for each range.

  3. Ongoing

    Live, as blocks commit

    The backfill hands over to live delivery on the same cursor, so there is no seam to reconcile.

Stop exporting. Start querying.

Kafka, Postgres, ClickHouse and S3 are included on Scale. Other destinations on request.

Get an API keyTalk to us